[gdal-dev] An issue about using GDALWARP to do theNTv2transformation

[Uwe Schmitz]

Heng,

heng.feng wrote:
> 
> Hi Uwe,
> 
> Thanks for your reply. The l0.tif file does not contain any georeferenced
> information. However, I have run the GDALINFO command on another raster
> data file which is in the same map data set provided by the customer and
> hence is supposed to be georeferenced in the way as the l0.tif file. The
> command output is sent as attached. As mentioned before, I also did the
> NTv2 transformation on  this raster data file
> (Notrufzonen_50m_v11_6_EPOS.tif) with the same GDALWARP command and the
> same grid shift file. I also got a lot of error messages saying that some
> coordinates are outside the area of the grid shift file. And with the
> "+x_0=500000" parameter, the output error messages are much less than the
> ones with the "+x_0=4500000" parameter.
> 
> Thanks for your help again.
> 

from your screenshot I see that georeference information 
of your file seems to be OK. So forget about my comment
about missing meridian prefixes, that's not the case.

But ..., your file is very large (the extent, not the size).
Please notice, that due to the nature of the NTv2 approach
it is not possible to calculate coordinates outside the
extent of the grid. The BeTA2007 grid covers longitudes
between 5d30' to 15d40' East. Your Screenshot of gdalinfo
shows that the longitude of the upper left corner is
4d54'45.24" East and hence outside the BeTA2007 grid.

OK, what can you do? BeTA2007 has submeter accuracy and
I believe, that you don't need it for you data (one pixel 
covers 50m*50m). So you can try the 7-Parameter Helmert
approach described by EPSG::1776 with "+towgs84=..."
Parameter instead of "+nadgrids=...". This will result in
errors of max. 3m in comparison to BeTA2007. Not that
bad for your data ;-)

Another way is to specify the "null" grid as fall back for 
coordinates outside the BeTA2007 grid, like so:

   ... +nadgrids=C:\fh\comar ch\BETA2007.gsb,null

But be aware that coordinates outside the BeTA2007 grid experience
no datum shift. This may result in errors up to 200m.

Hope that helps.

Regards
Uwe
-- 
View this message in context: http://osgeo-org.1803224.n2.nabble.com/An-issue-about-using-GDALWARP-to-do-the-NTv2-transformation-tp5838904p5862681.html
Sent from the GDAL - Dev mailing list archive at Nabble.com.